collaborative sensing for symbiotic human-robot environments
mel siegel intelligent sensors measurement & control lab the robotics institute – school of computer science carnegie mellon university – pittsburgh pa 15213
Robotic and Sensors Environments (ROSE-2006) Orebro University – Orebro SE 2006 May 31
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 1
disclaimers
• warning 1: this is a 100% data-free presentation intended only to identify issues, stimulate discussion, and suggest a systematic approach to resolving them • warning 2: when they dabble in human and social issues, physical scientists and engineers tend to forget everything they ever knew – and preached – about how important it is to make plans and decisions based on evidence not intuition (prejudice)
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 2
“chaos in the brickyard”
• a “fairy tale” explaining how we became confused about the difference between making “bricks” and building “edifices”
– Bernard K. Forscher (Mayo Clinic) Science 142:3590, October 18, 1963
http://www.health.uottawa.ca/biomech/csb/archives/chaos.htm
facts & gadgets
• Poincare, Forscher, and others (including I. I. Rabi and L. Lederman) who have said the same thing using slightly different metaphors were warning about our confusing of facts with the underlying laws • in the 40+ years since Forscher’s article we have seen the appearance of a parallel pernicious trend: we confuse building gadgets and “infrastructure” with the practical solution of real problems
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 4
• Le savant doit ordonner; on fait la science avec des faits comme une maison avec des pierres; mais une accumulation de faits n'est pas plus une science qu'un tas de pierres n'est une maison. – Henri Poincare
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 3
1
consequence
• “systems” with all the “infrastructure” and none of the sensors to make it work • • • • • • •
outline
introduction helping people sense robots helping robots sense people helping overseers sense both inherently-safe systems what do we need? conclusions and future prospects
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
5
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
6
contexts
• humans and robots working together in
– home – medical care – manufacturing – mining, construction, etc, outdoors – rescue – military – space and extraterrestrial planets, etc
introduction
environments
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 7 ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 8
2
requirements
• for safety
– enhanced levels of situational awareness
• by the human collaborator of the robot • of the human collaborator by the robot
realized in part via ...
• substantially improved capabilities for
– sensing
• of the robot by the human • of the human by the robot • of the robot and the human by the “overseer” (i.e., an environmentally-sited sensing, communication, and control entity)
• for effectiveness
– enhanced levels of coordination – enhanced levels of task-related anticipation
• of the robot of the human • of the human of the robot
– display
• of pose, dynamics, task-status, next-steps, etc • for the human collaborator and external monitors
for the collaborative task to succeed
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 9
• and the digital glue to hold it all together ...
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 10
digital glue
• commitment to the principle that it is unreasonable to expect the human to be the responsibly-behaving partner • analysis of how collaborations go wrong
– and what sensing capabilities will prevent them
helping people sense robots
• reasonable cost-benefit models
– is it reasonable – and ethical – to accept the cost to individuals if the benefit to society seems to outweigh it statistically ?
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
11
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
12
3
human senses are aided
human senses are augmented
http://gizmodo.com/gadgets/rearview-helmet
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 13
http://www.tekgear.com/index.cfm?pageID=90&prodid=385§ion=84&nodelist=1,83,84
http://www.gizmodo.com/gadgets/digital-cameras/headmounted-police-dog-camera-033871.php ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 14
human senses are exploited
robot signatures
natural signature of robots naturally sensed by humans human needs prosthesis • colors • sounds • air movement • metal • E&M fields • temperature signature easily added to robots • flashing lights • weird sounds • visual identifiers • magnets • RF/US xmtrs • -radiation (?)
16
sarkar / vanderbilt university
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 15 ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS
4
sensor friendliness
• concept emerged in context of automated or automation-aided driving • friendly roadway and/or friendly vehicle are equipped with (simple, inexpensive) tags that make them easier to sense
– license plate w/ augmented radar cross-section – RF transponders – bar magnets embedded in the roadway – fluorescent paint coding lane dividing lines – tail-lights modulated to show driver’s intentions
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 17 ROSE-2006 05 31
sensor friendly robots
• many of the concepts that arose in the context of sensor friendly vehicles and roadways are portable to the human-robot collaboration context
mws@cmu.edu ISMCL RI SCS
18
robot actuators are exploited
helping robots sense people
kosuge & hirata / tohoku university
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 19 ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 20
5
human signatures
natural signature of humans naturally sensed by robots robot needs prosthesis
ROSE-2006 05 31
whiskers
• mechanical? fiber-optic? piezoelectric? etc
signature easily added to humans
• none before it is too late? • vision • hearing • temperature • whiskers?
• none?
• special clothing • magnets • RF/US xmtrs
21 ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 22
mws@cmu.edu ISMCL RI SCS
sensor friendly humans??
• theorem: no, it is unreasonable to expect humans to be the parties responsible for safety in human-robot interactions • lemma 1: okay, okay, maybe for some kinds of collaborative work between a robot and a trained professional worker we can assign some responsibility to the human • lemma 2: but never, never, when the interaction is between a robot and a “lay person”
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 23 ROSE-2006 05 31
helping overseers sense both
mws@cmu.edu ISMCL RI SCS
24
6
environment is exploited
joint signatures
natural signature of human & robot naturally sensed by overseer overseer needs prosthesis signature easily added to human & robot
• visible presence • distinctive colors (assuming • unusual patterns ubiquitous cameras) • sounds of work • distinctive lights • warmth • RF/US xmtrs • E &/or M fields • E, M or EM fields
mws@cmu.edu ISMCL RI SCS 26
mizuguchi / tokyo university
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 25 ROSE-2006 05 31
issues
• privacy • privacy • privacy
inherently-safe systems
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
27
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
28
7
issues
• what/how can we learn from efforts that have been made to make engineered systems “inherently safe”, e.g., nuclear reactor designs that are guaranteed to shut down if/when their control systems fail?
means ...
• if a component fails ... • or if the control system shuts down ... • the system evolves into a safe steady state
– people are not hurt
• operators • subjects (rescuees, patients, passengers, etc) • by-standers
– the mission succeeds (or does not fail) – the robot is not damaged
• (application-dependent priorities of last two)
ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 29 ROSE-2006 05 31 mws@cmu.edu ISMCL RI SCS 30
discussion ...
• ... by all ...
what do we need?
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
31
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
32
8
discussion ...
conclusions and future issues
• ... by all ...
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
33
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
34
that’s all folks: mushroom medley
thanks for your participation ...
• my contact information: mel siegel mws@cmu.edu +1 412 268 8742 office/lab http://www-2.cs.cmu.edu/~mws
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
35
ROSE-2006 05 31
mws@cmu.edu ISMCL RI SCS
36
9